Fed Chair Kevin Warsh and the FOMC Just Hiked Interest Rates, and 36 Years of History Make Clear What Comes Next for Stocks

The FOMC raised the federal-funds target rate by 25bps to 3.75%-4.00% on Sept. 16, the first increase since July 2023, and its projections indicate another 25bp hike before year-end. The Dow fell more than 1% following the decision, while the S&P 500 and Nasdaq also declined, as Chair Kevin Warsh stressed a faster return to the Fed's 2% inflation target. Despite near-term valuation and AI infrastructure-financing risks, historical data cited in the article show the S&P 500 was higher 12 months after every prior 25bp initial hike since 1990, averaging a 12.5% gain, though it was lower one month later in all cases.

Analysis

The relevant transmission mechanism is not the policy-rate level alone but the curve, credit spreads, and refinancing calendar. Cash-rich hyperscalers can sustain AI capex from operating cash flow, limiting direct sensitivity for NVDA over the next 1-3 quarters; the more vulnerable nodes are leveraged data-center landlords (DLR, EQIX), merchant compute providers, and lower-quality power/thermal suppliers whose valuations assume cheap external capital. A higher-for-longer path also raises the hurdle rate for marginal AI projects, shifting spend toward the highest-utilization GPU clusters and strengthening NVDA's competitive position versus less differentiated infrastructure vendors.

The historical first-hike comparison is weak evidence: prior cycles began from materially different inflation, fiscal, valuation, and index-concentration regimes. Near term, systematic de-risking can pressure long-duration software and AI beneficiaries even if growth remains intact; over 1-3 months, the decisive catalyst is whether inflation and payroll data permit the expected terminal-rate path to stabilize. Over 6-18 months, persistent restrictive policy would likely favor profitable platform companies and financials over capital-intensive real estate and unprofitable software.

Contrarian view: consensus may overstate NVDA's rate exposure while understating the vulnerability of AI-adjacent real assets. The key risk to an AI-duration short is a rapid decline in inflation that compresses real yields, reopens project-finance markets, and restores the premium paid for data-center capacity. Conversely, a widening in BBB spreads or material downward revisions to hyperscaler capex would invalidate the assumption that AI demand can absorb incremental supply.

Key Decisions for Investors

• Initiate a 3-6 month pair: long NVDA / short DLR, sized beta-neutral. NVDA has superior earnings visibility and customer balance-sheet support; DLR carries greater refinancing and cap-rate sensitivity. Reassess if 10-year real yields fall more than 50 bps or if DLR raises stabilized-development guidance.
• Overweight XLF versus IGV for the next 1-3 months. Banks and insurers benefit if the curve steepens without a credit event, while software multiples remain exposed to discount-rate repricing. Stop the pair if high-yield spreads widen above roughly 450 bps, signaling that credit losses rather than net-interest-income tailwinds are becoming dominant.
• Do not add broad AI-infrastructure shorts solely on rate fears. Set an alert for quarterly hyperscaler capex guidance and NVDA data-center revenue expectations; a coordinated cut in capex plans or NVDA guide below consensus is the confirmation needed for a tactical short in SMH or SOXX.
• Maintain only a modest tactical underweight in NFLX rather than a directional short: its subscription cash flows are relatively durable, but its valuation remains duration-sensitive. Cover on a material reacceleration in net adds or if real yields retrace below pre-hike levels.